9374 Phys. Chem. Chem. Phys., 2011, 13, 9374–9384 This journal is c the Owner Societies 2011
Cite this: Phys. Chem. Chem. Phys., 2011, 13, 9374–9384
Molecular dynamic simulation of dicarboxylic acid coated aqueous
aerosol: structure and processing of water vapor
Xiaofei Ma,aPurnendu Chakraborty,
aBrian J. Henz
cand Michael R. Zachariah*
ab
Received 23rd September 2010, Accepted 15th March 2011
DOI: 10.1039/c0cp01923b
Organic monolayers at the surfaces of aqueous aerosols play an important role in determining the
mass, heat transfer rate and surface reactivity of atmospheric aerosols. They can potentially
contribute to the formation of cloud condensation nuclei (CCN) and are involved in a series
of chemical reactions occurring in atmosphere. Recent studies even suggest that organic-coated
interfaces could have played some role in prebiotic biochemistry and the origin of life. However,
creating reproducible, well-characterized aqueous aerosol particles coated with organic films is
an experimental challenge. This opens the opportunity for computer simulations and modeling
of these complex structures. In this work, molecular dynamics simulation was used to probe
the structure and the interfacial properties of the dicarboxylic acid coated aqueous aerosol.
Low molecular weight dicarboxylic acids of various chain lengths and water solubility were
chosen to coat a water droplet consisting of 2440 water molecules. For malonic acid coated
aerosol, the surface acid molecules dissolved into the water core and formed an ordered structure
due to the hydrophobic interactions. The acid and the water are separated inside the aerosol.
For other nanoaerosols coated with low solubility acids, phase separation between water and
acid molecules was observed on the surface of the particle. To study the water processing
of the coated aerosols, the water vapor accommodation factors were calculated.
Introduction
Organic material is ubiquitous in the earth’s atmosphere and
represents an important fraction of the fine aerosol mass.
Studies have shown that total organic carbon can represent
10–65% of the aerosol mass and exists as a complex mixture of
hundreds of organic compounds, while secondary organic
carbon can contribute up to 25–50% of the fine aerosol mass
in urban polluted areas.1 Indirectly, atmospheric aerosols can
affect the radiative properties and lifetime of clouds and thus
have an influence on global climate by acting as cloud
condensation nuclei (CCN).2 Observations have revealed that
more than 60% of the CCN can consist of organic constituents.3
Recent experimental studies and thermodynamic analysis of
organic marine aerosols even suggest that atmospheric aerosols
could act as prebiotic chemical reactors and play a role in the
origin of life.4,5 Despite the considerable fraction of organic
matters in atmospheric aerosols and significant importance of
their environmental and biological functions, little is known
about their structure and influence on atmospheric processes.
The organic materials can be water-soluble and insoluble,
volatile and nonvolatile, surface-active and surface-inactive,
and biogenic and anthropogenic. Depending on their physical
properties (e.g. solubility and volatility), the organics can form
different structured films on existing aerosol particle surfaces.
Water-insoluble organic molecules are likely to be closed-packed
and oriented and thus tend to form ‘‘condensed films’’ on the
particle surfaces. Phase transitions which correspond to differing
degrees of ordering of the surfactant molecules can take place
in those films.6 Our previous molecular dynamic simulation
results7,8 on the structure of long-chain fatty acid coated
nanoaerosols showed that in the final stage of equilibrium,
an inverted micelle structure is formed, and consistent with
a previously proposed ‘‘inverted micelle’’ model.9 In this
structure a water core is surrounded by surface adsorbed fatty
acid molecules. On the other hand, water-soluble organic
surfactant molecules tend to form less compact films which
do not undergo phase transitions to more compact structures.
The primary relevance of these structures is how they
subsequently interact with other organics, accommodation of
water vapor, and its ability to act as cloud condensation nuclei.
There is experimental evidence that organic compounds perturb
the uptake of trace gases onto aqueous surfaces.10 The presence
of the organic films on water drops could significantly alter both
condensation and evaporation rates. To complicate this already
difficult problem, atmospheric ‘‘processing’’ of the surface by
aDepartment of Mechanical Engineering and Department ofChemistry and Biochemistry, University of Maryland, College Park,MD 20742, USA. E-mail: [email protected]
bNational Institute of Standards and Technology, Gaithersburg,MD 20899, USA
cU.S. Army Research Laboratory, Aberdeen Proving Ground,MD 21005, USA
PCCP Dynamic Article Links
www.rsc.org/pccp PAPER
This journal is c the Owner Societies 2011 Phys. Chem. Chem. Phys., 2011, 13, 9374–9384 9375
atmospheric oxidants would further alter the surface properties
of the aerosol, leading to further changes in reactivity.6,9
Among the various kinds of organic compounds, low
molecular weight dicarboxylic acids have attracted much
attention due to their large prevalence and interesting physico-
chemical properties. These types of acids have been identified
as one of the major organics in both urban and rural areas and
are a ubiquitous organic aerosol constituent in the marine and
even Arctic atmosphere.11,12 Observations have shown that
dicarboxylic acids are also commonly found in the organic
fraction of secondary aerosols. However, their formation and
partition to the aerosol phase are still unclear. In biology,
dicarboxylic acids are important metabolic products of fatty
acids. During recent years, a considerable effort has been
made to understand the properties of low molecular weight
dicarboxylic acids (C3–C9). It is known that the physico-
chemical properties of low molecular weight dicarboxylic acids
such as solubility, vapor pressure, evaporation rate, melting
and boiling points alternate with the number of carbon
atoms.13,14 Those physicochemical properties have profound
effect on the CCN activity. One of the major questions
surrounding organic compounds focuses on the changes in
surface tension of the droplet due to the presence of the
organic and the solubility of the compound.2 Water-soluble
materials are known to affect droplet activation by lowering
the surface tension and thus changing the critical droplet
radius. Experiments have confirmed that this effect can be
well predicted by the Kohler theory for soluble inorganic
species and organics that are wettable by water,15 when the
two components are homogenously dispersed. However,
when considering extending the current theory, low-solubility
organic species are equally important. The low molecular
weight dicarboxylic acids (C3–C9) cover a wide range of
solubility and thus provide an excellent platform to study
the effect of solubility on the CCN activation.
With hydrophilic groups at both ends of a hydrophobic
hydrocarbon chain, dicarboxylic acids are bolaamphiphilic
molecules. The structure and phase behavior of these
molecules in a particular type of medium are determined by
unique intermolecular interactions: the hydrophobic inter-
actions between hydrocarbon chains, the hydrophilic and/or
the electrostatic interactions between the head groups. When
amphiphilic molecules are dispersed in water, the hydrophobic
interactions of the hydrocarbon chains drive the molecules to
self-assemble into structures where the hydrophobic tails are
shielded from unfavorable interactions with water by the
hydrophilic, polar head groups.16 Like amphiphilic molecules,
bolaamphiphilic molecule aggregation is driven by hydrophobic
interactions, which can form self-assembly structures, including,
spherical lipid particles, vesicles produced from long-chain
molecules, and micelles from short-chain, water-soluble
bolaamphiphiles.17 Not surprisingly compared with amphilic
molecules, the introduction of a second hydrophilic head
group generally induces a higher solubility in water and an
increase in the critical micelle concentration.
Previous experiments have been carried out to investigate
the cloud activity of various pure dicarboxylic acid aerosols
from highly-soluble acids to almost insoluble acids.11,15,18 The
TDMA (Tandem DMA) method is frequently used in
laboratory studies. In this method, the first DMA produces
nearly monodisperse particles of a known size while the second
DMA measures the particle size distribution of the final
aerosol. However, there are some drawbacks associated with
this method. First, the DMAs are designed for classifying
spherical particles, and the results are interpreted based on a
singly charged assumption. Therefore, depending on particle
morphology, mass and cross-sectional area which affects the
charging efficiency, there is a possibility that DMA could
lead to incorrect size classification. Second, the ion-mobility
method deals with the whole aerosol population, so it is
impossible to use this method to monitor the water processing
of individual particles. In fact, since molecular processes
involve dynamics happening over short distances (nanometer
length scale) and short times (nanoseconds time scale), these
processes are difficult to probe experimentally. Furthermore,
the atmospheric organic aerosols usually consist of more than
one chemical species, and the organics can form complex
structures such as monolayers, thin films on the aerosol
surfaces. It is an experimental challenge to create reproducible,
well-characterized aqueous aerosol particles which are coated
with an organic film. However, on the other hand, the
structure of complex aerosols can be explicitly defined in
molecular simulations and the dynamics of the molecular-
scale process can be followed explicitly as well.
In this work, we employed a molecular dynamic simulation
method to study the structure evolution and water processing
of various dicarboxylic acids coated water droplets. Two
questions are addressed in our study: (1) What is the relation-
ship between the aerosol final structure and physical properties
of coated dicarboxylic acid molecules? (2) How does the
aerosol structure affect the water processing of the coated
aerosol?
Computational model and simulation details
The molecular dynamics simulations in this work were carried
out using Lammps19 (Large-scale Atomic/Molecular Massively
Parallel Simulator, http://lammps.sandia.gov) a software
package developed by Sandia National Laboratories.
1 Simulating the water core
For simulating the water molecule, we used the extended
simple point charge (SPC/E) interaction potential developed
by Berendsen et al.20,21 This model consists of a total of three
sites for the electrostatic interactions of water with an OH
distance of 0.1 nm and HOH angle of 109.471. The partial
positive charges of +0.4328e on the hydrogen atoms are
exactly balanced by partial negative charges of �0.8476elocated on the oxygen atom. A pairwise Lennard-Jones inter-
action centered on the oxygen atoms is used to compute
the van der Waals forces between two water molecules. The
expression is given by:
VLJðrijÞ ¼ �A
rij
� �6
þ B
rij
� �12
ð1Þ
withA=0.37122 (kJmol�1)1/6 nm,B=0.3428 (kJmol�1)1/12 nm,
where rij is the distance between two oxygen atoms. The SPC/E
9376 Phys. Chem. Chem. Phys., 2011, 13, 9374–9384 This journal is c the Owner Societies 2011
potential has been studied extensively. It can provide accurate
reproduction of water surface tension and its temperature
dependence22 and the effects of simulation size and the treat-
ment of long-range interactions on surface tension have been
shown to be less than 10%.23
The first step toward building a dicarboxylic acid coated
aqueous aerosol was to build a pure water droplet.
The detailed procedures for preparing an equilibrated water
droplet are described in our previous work.7 In brief, the water
droplet is evolved from a simple cubic lattice structure with
the oxygen atom at the vertex of each cube. The initial
sphere configuration consisting of 2440 water molecules was
generated by considering only the water molecules inside a
sphere of a certain radius. The water droplet is equilibrated at
the temperature of 260 K. The SHAKE algorithm24 was used
to apply constraints between the O–H bond and H–O–H angle
to maintain the rigidity of the water molecules. The radial
distribution function confirms the liquid phase of the prepared
water droplet.
2 Modeling the dicarboxylic acid molecule
The next step was to coat the spherical water droplet with
dicarboxylic acid molecules. By identifying the surface molecules
of the water droplet, we placed dicarboxylic acid molecules on
the surface water sites with one of the carboxyl groups of the
dicarboxylic acid attached to the surface water molecule and the
straight chain of the dicarboxylic acid placed radially outward.
The initial coating density is about 6.6 molecules per nm2.
The dicarboxylic acids simulated were malonic acid (C3),
succinic acid (C4), glutaric acid (C5), adipic acid (C6), pimelic
acid (C7), suberic acid (C8), azelaic acid (C9) and branched
azelaic acid (C9_branched). The physical properties of these
dicarboxylic acids are summarized in Table 1. For modeling
the dicarboxylic acid molecule, we used a mixture of the
‘‘fully atomistic’’ method and the ‘‘united atom’’ method.
The high-accuracy/high computational requirement fully
atomistic method is used to model the carboxyl group
(–COOH group) while the less accurate/lower computational
requirement united atom method is employed to simulate the
methylene group (–CH2 group). In the united atom setup, the
number of interaction sites is reduced by subsuming some or
all of the atoms into the atoms to which they are bonded.
Thus considerable computation savings are possible. In our
simulation, each methylene group (–CH2 group) is represented
by a single site with interactions defined between these sites.
An example of the dicarboxylic acid in this work is shown in
Fig. 1. During the simulation, the O–H bond in the carboxyl
group was kept rigid using the SHAKE algorithm.
The reliability of predictions from molecular simulations is
determined largely by the accuracy of the representation of the
intermolecular interaction potentials. The potentials used in
our model include nonbonded interactions between each pair
of atoms and bonded interactions between bonded atoms
which have contributions from bond stretching, angle vibra-
tion, proper and improper dihedral interactions. For non-
bonded force calculation, the switched Lennard-Jones and
the switched coulombic potential were used to simulate the
van der Waals interactions and the electrostatic interactions,
respectively. The switched forces are given by:
E ¼
LJðrÞ rorin
SðrÞ � LJðrÞ rinororout
0 r4rout
8>>><>>>:
E ¼
CðrÞ rorin
SðrÞ � CðrÞ rinororout
0 r4rout
8>>><>>>:
LJðrÞ ¼ 4esr
�12� s
r
�� 6� ��
CðrÞ ¼ Cqiqj
er
SðrÞ ¼ ½r2out � r2�2½r2out þ 2r2 � 3r2in�
½r2out � r2in�3
ð2Þ
Harmonic potentials of the form E= K(x � x0)2 (where K is a
prefactor, x is the position vector or angle and x0 is the
corresponding equilibrium value) were chosen to model
bond stretching and angle vibrations. The periodic function
E = K[1 + cos(nf � d)] was used to describe the proper
dihedral interactions. For the improper interactions, a harmonic
potential E = K(w � w0)2 was chosen to keep a planar group
in the same plane. The potential parameters used in these
simulations were obtained from the Gromacs force field
database and are listed in Table 2.
Table 1 Physical properties of the dicarboxylic acids studied in this work
Dicarboxylic acid Chemical formula Molar mass/g mol�1 Density/g cm�3 2,11,14Solubility in water/g per 100 g water2,11
Malonic acid (C3) HOOC–(CH2)–COOH 104.06 1.631 161Succinic acid (C4) HOOC–(CH2)2–COOH 118.09 1.572 8.8Glutaric acid (C5) HOOC–(CH2)3–COOH 132.12 1.424 116Adipic acid (C6) HOOC–(CH2)4–COOH 146.14 1.360 2.5Pimelic acid (C7) HOOC–(CH2)5–COOH 160.17 1.281 71Suberic acid (C8) HOOC–(CH2)6–COOH 174.20 1.272 0Azelaic acid (C9) HOOC–(CH2)7–COOH 188.22 1.251 0
Fig. 1 Structure of a dicarboxylic acid (C6).
This journal is c the Owner Societies 2011 Phys. Chem. Chem. Phys., 2011, 13, 9374–9384 9377
3 Equilibration procedures
After attaching a monolayer dicarboxylic acid to the surface of
the water droplet, an energy minimization run was performed
using the conjugate gradient (CG) algorithm to relax the initial
configuration containing highly overlapped atoms. The system
was then allowed to equilibrate at 0 K for 10 ps using constant
NVE integration with the velocity-Verlet algorithm to update
position and velocity for all atoms. A timestep of 1 fs was
typically chosen to ensure energy conservation. During the
equilibration process, the temperature was controlled by rescaling
the velocities every timestep. After the system relaxation at
0 K, the coated particle was slowly heated to 260 K over 200 ps
and then allowed to equilibrate for 400 ps. Following that, the
system was heated to 300 K and allowed to equilibrate at that
temperature for up to 8 ns. For the final step of the prepara-
tion process, the simulations were switched from a constant
temperature to a constant energy calculation to ensure that the
average system temperature did not deviate by more than 10 K.
The simulations were carried out in a constant energy environ-
ment during the data generation phase. All the simulations
were run on a Linux cluster, running in parallel on 8 processors.
Our mixture method is computationally more efficient than the
fully atomistic method yet provides sufficient accuracy for the
analysis.
Results and discussion
In this section, the results from the molecular dynamics
simulations are presented and discussed. In order to study
the effect of dicarboxylic acid chain length on the structure of
coated nanoaerosols, we computed several physical properties,
including the radial density distributions, diffusion coefficients
and radial distribution functions. In order to study water
vapor processing of the coated nanoaerosols, water molecules
were impinged on the surface with a Boltzmann energy
distribution to obtain thermal sticking coefficients.
1 Structure
The time evolution of different dicarboxylic acids coated
nanoaerosol structures were monitored during the simulation
process. Different equilibrium morphologies were observed.
For C3-coated nanoaerosol, the C3 molecules slowly migrate
from the surface to the core of the nanoaerosol. This seems to
occur in two states. In the first B1 ns, the C3 molecules
dissolve and are dispersed inside the water droplet. By 5–6 ns,
however the dispersed C3 molecules dissolve further into the
water core and assemble around the center of the nanoaerosol.
The initial and final morphologies of C3-coated nanoaerosol
are presented in Fig. 2(a).
The evolution paths of C5–C9 coated nanoaerosols are
essentially similar to each other but significantly different from
that of C3-coated nanoaerosol. As a representative system, we
present the initial and final morphologies of C8-coated nano-
aerosol in Fig. 2(b). Unlike C3 molecules, C8 molecules
cannot dissolve into the water droplet. The result is phase
separation leading to the acid molecules segregation to one
side of the droplet, and forming a multilayer structure, similar
to what might be expected for the separation of oil and water.
This behavior was observed for C5–C9 coated nanoaerosols.
Apparently, for sufficiently long molecules the combined
effects of the acid–acid interaction with the hydrocarbon
backbone interactions are more competitive than the acid–
water interaction. The equilibrated structure of C4-coated
nanoaerosol is found to intermediate between the C3 and C8
cases with some of the acid molecules dissolved into the water
core while the rest formed a separate acid cluster.
To investigate the effect of a branched chain on the struc-
ture evolution of dicarboxylic acid coated nanoaerosol, a
C9 molecule was modified to add two side CH3 groups.
Fig. 2(c) shows the initial and final structures of the system.
A schematic of the branched acid structure is also shown in
Fig. 2(c). Different from previous cases, the C9_branched acid
molecules did not dissolve into the water droplet or form a
separate acid cluster, but rather maintained the initial inverted
micelle structure. Apparently, the addition of a side group
actually stabilizes the monolayer, by constraining the molecular
conformation to a linear structure.
Table 2
(1) Nonbonded force parameters
Atom e/kcal mol�1 s/A
OW–OW 0.1553 3.166HW1–HW1 0.0 0.0HW2–HW2 0.0 0.0HO–HO 0.0 0.0OA–OA 0.2029 2.955C–C 0.0970 3.361O–O 0.4122 2.6260CH2–CH2 0.1400 3.9647
(2) Harmonic bond parameters
Bond K/kcal mol�1 A�2 r0/A
OW–HW1; OW–HW2 0.0 1.0HO–OA 375.0 1.0OA–C 450.0 1.36CQO 600.0 1.23C–CH2; CH2–CH2 400.0 1.53
(3) Harmonic angle parameters
Angle K/kcal mol�1 rad�2 y0/1
HW1–OW–HW2 0.0 109.47HO–OA–C 47.5 109.5OA–CQO 60.0 124.0OA–C–CH2 60.0 115.0OQC–CH2 60.0 121.0C–CH2–CH2; CH2–CH2–CH2 55.0 111.0
(4) Proper dihedral parameters
Dihedral K/cal mol�1 n (integer) d/1
HO–OA–C–CH2; HO–OA–CQO 4.0 2 180OA–C–CH2–CH2; OQC–CH2–CH2 0.1 6 0C–CH2–CH2–CH2; CH2–CH2–CH2–CH2 1.4 3 0
(5) Improper dihedral parameters
Improper dihedral K/cal mol�1 rad�2 w0/1
C OA CH2 O 40 0.0
9378 Phys. Chem. Chem. Phys., 2011, 13, 9374–9384 This journal is c the Owner Societies 2011
2 Radial density
The density profiles of the dicarboxylic acid coated nanoaerosols
were calculated as a function of radial distance r. To calculate
density, we considered a shell of thickness dr at a distance
r from the center of mass. The density at a radial distance r is
given by the mass of all the sites in that shell divided by the
volume of the shell. Due to the lack of spherical symmetry,
the density calculation could not be performed on C4–C9
dicarboxylic acid coated nanoaerosols. The calculated density
profiles for C3 and C9_branched dicarboxylic acid coated
nanoaerosols are presented in Fig. 3.
The density profiles for the C3 acid coated nanoaerosol
show, in Fig. 3(a), the acid density distribution gradually
shifting to the inside of the aerosol. After about 5–6 ns of equili-
bration, both water and acid density profiles converge to their
equilibrium values. Curiously the water density shows a bimodal
distribution, but overall the structure would indicate that the
surface is enriched in water relative to the interior of the droplet.
On the other hand the density profiles of C9_branched acid
coated nanoaerosol (Fig. 3(b)) maintain a core-shell structure.
3 Radial distribution function
The radial distribution function (RDF), g(r), is defined as the
number of atoms a distance r from a given atom compared
with the number of atoms at the same distance in an ideal
gas. In our simulation, the RDF for each atom pair is
histogrammed into 100 bins from distance 0 to the maximum
force cutoff distance. To investigate the microstructure
changes of the dicarboxylic acid coated nanoaerosols, RDFs
between different atom pairs were calculated. The RDF of the
water oxygen–oxygen pair is a good indicator of changes in the
water structure and RDF for water oxygen and the carbon in
the acid head group, tracks the interaction between water and
acid. Finally the RDF of the acids central CH2–CH2 groups
characterizes the interaction between dicarboxylic acid
molecules.
Fig. 4(a) shows the water oxygen–oxygen RDFs for C3, C8
and C9_branched acid coated nanoaerosols after equilibra-
tion. For all three RDFs, the major peak occurs at roughly
2.8 A which is the well-known average hydrogen bond length
in water. Fig. 4(b) compares the RDFs of the central
CH2–CH2 groups for the three representative nanoaerosols
after equilibration. The RDF for the C3 case exhibits a well
resolved peak at around 5 A, while the RDFs for C8 and
C9_branched chains are much more diffuse. The reader is
reminded that in the final equilibrium stage, the coating
C3 acid molecules all dissolved into the water core. Thus, a
clear peak in the RDF indicates that the C3 acid molecules
apparently form a short-range ordered structure within the
water core. The time evolution of the RDFs of acid central
CH2–CH2 in the C3 case is presented in Fig. 5(a). The system
starts from an initial ‘‘ordered structure’’ where all the acid
molecules are pointing radially out of the water droplet, after
1 ns of equilibration, the acid molecules dissolve into the water
droplet. This leads to the decrease in the peaks of RDF at 1 ns.
As the system continually evolves, the peaks in RDFs
grow which indicates the formation of a more ordered acid
structure in water. To investigate the relative orientation
Fig. 2 Structures of dicarboxylic acid coated nanoaerosols: (a) struc-
ture of C3 coated nanoaerosol, (b) structure of C8 coated nanoaerosol,
(c) structure of C9_branched coated nanoaerosol (c).
This journal is c the Owner Societies 2011 Phys. Chem. Chem. Phys., 2011, 13, 9374–9384 9379
between the C3 acid molecules, RDFs of different groups
within the acid molecule are calculated both for the initial
configuration and for the equilibrated structure. The results
are presented in Fig. 5(b) (initial structure) and Fig. 5(c)
(equilibrated structure). In Fig. 5(b), groups 1, 2 and 3
represent the carbon atom in the head group, CH2 group,
and the carbon atom in the tail group for a C3 molecule,
respectively. As we can see from Fig. 5(b), the RDF for group
1 atoms has the highest peak intensity, while the RDF for
group 3 atoms has the lowest peak intensity, which reflects the
initial ‘‘radially outward’’ molecular configurations of C3
molecules. After equilibration, the RDFs for group 1 atoms
and group 3 atoms almost overlap which indicates a equal
separation between the group 1 atoms and group 3 atoms. The
RDF for group 2 atoms peaks at a slightly large separation.
This suggests a pairwise parallel configuration between C3
molecules (as shown in the inset of Fig. 5(c)) with the separa-
tion between the central CH2 groups slightly larger than that
for the head and tail groups.
To investigate the phase separation occurring in long chain
dicarboxylic acids coated nanoaerosols, the RDFs of the atom
pair between the water oxygen and the carbon atom in the acid
head group for the C8 case were calculated at different stages of
equilibration. The results are presented in Fig. 6. As we can see
from the figure, starting from the initial ‘‘ordered configuration’’,
the peaks in the RDFs continuously decrease, indicating the
increased separation between water and acid molecules.
4 Phase separation mechanism
In order to study the phase separation of the long chain acids
in more detail, a low surface coverage (0.6 molecules per nm2),
C8 coated nanoaerosol was prepared and allowed to evolve
from an initial ‘inverted micelle’ configuration. The images of
structure evolution of this nanoaerosol are presented in Fig. 7.
As can be seen from Fig. 7, the phase separation process
between the acid and water molecules occurs in several stages.
In the first stage, the C8 dicarboxylic acid molecules flop
down onto the water droplet surface to maximize interaction
between the two acid groups and water (0.1 ns–1 ns). The next
stage of the phase separation process is the aggregation of the
acid molecules. Here the dicarboxylic acid surface diffuses and
aggregates to form multi-layered clusters of acid molecules
which favor acid–acid interactions (1 ns–2.5 ns). In turn these
clusters also surface diffuse to aggregate into larger clusters.
5 Discussion: hydrophobicity and the structure of dicarboxylic
acid coated aqueous aerosol
The hydrophobic effect is the tendency for non-polar molecules
and water to segregate, and is an important driving force for
Fig. 3 Radial density distributions of (a) C3 and (b) C9_branched acids coated nanoaerosols as a function of simulation time.
9380 Phys. Chem. Chem. Phys., 2011, 13, 9374–9384 This journal is c the Owner Societies 2011
amphiphlies self-assembly and biological folding.25 Trying to
mix hydrophobic molecules with water requires a decrease in
hydrogen bonding, which if not energetically favored, just
results in segregation of non-polar molecules from water, and
an effective attraction between hydrophobic molecules.26 If the
hydrophobic molecule is small enough, such that the disrup-
tion of the hydrogen bond network is also small, then water
molecules can adopt orientations to go around the small
hydrophobic solute without sacrificing hydrogen bonds,
or move away from the large hydrophobic object and
form an interface around it.27 The critical length scale dividing
large hydrophobic units from small has been estimated to
be B1 nm.27
The dicarboxylic acid molecules studied in this work contain
a hydrophobic hydrocarbon chain, which can participate in
the hydrophobic interaction with water. The hydrophilic parts
of the dicarboxylic acid molecules are not directly responsible
for hydrophobic assemblies, but they can affect the arrangement
of these assemblies relative to interfaces and other structures. The
structures observed in our simulation for different dicarboxylic
acid coated nanoaerosols can be explained successfully using the
knowledge of hydrophobic interactions between the acid mole-
cules and the water molecules.
Fig. 4 RDFs for different atom pairs for C3, C8 and C9_branched
dicarboxylic acid coated nanoaerosols. (a) RDFs of water O–water O
and (b) RDFs of acid central CH2–CH2 groups.
Fig. 5 (a) Time evolution of the RDFs of acid central CH2–CH2 in
the C3 case, (b) RDFs of atom groups within acid molecules at the
initial configuration; the inset shows the grouping of atoms for a
C3 molecule, (c) RDFs of atom groups within acid molecules after the
equilibration process; the inset shows a cartoon of the pairwise parallel
configuration between C3 molecules.
This journal is c the Owner Societies 2011 Phys. Chem. Chem. Phys., 2011, 13, 9374–9384 9381
For C3 acid coated nanoaerosol, we observed a final
structure with the surface acid molecules completely dissolved
within the water core to form an ordered cluster (see g(r) in
Fig. 4(b)) at the center of the water droplet. As a result, the
original water molecules at the droplet center were excluded,
leading to a very low water density at the center (see Fig. 3(a)).
This structure is readily understood in terms of the dependence
of hydrophobic solvation on the size of hydrophobic unit.
When the C3 acid molecules cluster together they form a
sufficiently large hydrophobic unit (i.e. large volume to surface
ratio), so that the salvation free energy of the acid cluster is
lower than the overall salvation free energy of the individual
acid molecules. By forming a nearly spherical acid cluster, the
total acid surface area exposed to the unfavorable hydrophobic
interactions with water is thus minimized. The diameter of this
acid cluster is about 2 nm which is consistent with the criteria
that above 1 nm the energetic cost of assembling hydrophobic
units is significantly more favorable than the entropic cost of
keeping them separate.25
For nanoaerosols coated with low solubility dicarboxylic
acids, we observed phase separation between acid and water at
the surface of the water droplet with the acid molecules
forming a layered aggregate. The radial distribution of the
acid aggregate reveals that this aggregate structure is less
ordered than the cluster structure formed by C3 acid molecules
in water. Since the free energetic cost of dissolving large
dicarboxylic acid molecules in water is formidable, the acid
molecules will stay on the water droplet surface. However, as
we can see from Fig. 6, due to the two hydrophilic head
groups, the first step towards acid molecule aggregation is that
each individual acid molecule folds on the surface of water
droplet. This configuration is not energetically favorable since
the hydrocarbon chain in each dicarboxylic acid molecule is
close to the water. The equilibration process is to minimize this
unfavorable interaction, by minimizing the total acid surface
area exposed to water, resulting in aggregation. Since this
aggregation process happens near the surface of water droplet,
without water medium around the acid aggregate the hydro-
phobic interaction is not as strong as that in water. Therefore,
the final structure of acid aggregate is less ordered than the
structure of the C3 acid cluster in water.
For the high surface coverage (6.6 molecules per nm2),
C9_branched acid coated nanoaerosol, with the addition of
a branched chain, the C9_branched acid molecule is hard to
fold completely on the water surface due to the geometric
constraints. Thus, the hydrophobic part of the acid molecule is
not affected by the unfavorable interaction with water. The
initial structure of C9_branched acid coated nanoaerosol
keeps intact. However, for a low surface coverage C9_branched
acid coated nanoaerosol, we do observe the phase separation
between the acid and the water molecules, since in this case the
acid molecules can fold on the surface of water droplet. Fig. 8
compares the initial and equilibrated configurations of a low
surface coverage (0.6 molecules per nm2) C9_branched acid
coated nanoaerosol. A phase separation can be clearly seen.
6 Diffusion
In order to gain a better understanding of the dicarboxylic
acid coating and how it interacts with the water droplet, we
computed the diffusion coefficients for the dicarboxylic acid
molecules and the water molecules. This is achieved by
computing the mean square displacement of the dicarboxylic
acid and water molecules in the simulation and evaluating the
diffusion coefficient using eqn (3):
@hr2ðtÞi@t
¼ 2dD ð3Þ
where hr2(t)i is the mean square displacement (MSD) of the
molecules being tracked, t is time, d is the dimension available
for diffusion (in our case 3), and D is the diffusion coefficient.
The calculated diffusion coefficients for different dicarboxylic
acid molecules are presented in Fig. 9 as a function of molecular
weight. The best-fit line of the calculated self-diffusion coefficients
shows a (molecular weight)�1 dependence, which is consistent
Fig. 6 RDFs of water O-acid head group C for C8 coated nanoaerosol
at different equilibration stages.
Fig. 7 Phase separation of low coverage (0.6 molecules per nm2), C8
acid coated nanoaerosol.
9382 Phys. Chem. Chem. Phys., 2011, 13, 9374–9384 This journal is c the Owner Societies 2011
with the Rouse model28 which makes a ‘‘no entanglements’’
assumption, which is likely true for the relatively small
acids being considered here. Interestingly the same molecular
weight dependence on the diffusion coefficient applies for both
the water soluble (C3 and C4) and non-soluble (C5–C9)
molecules. The reader is reminded that the C3 acid molecules
are fully dissolved into the water core and the C4 acid
molecules are partially dissolved into the water. While for
the longer chain acid molecules, a close-packed cluster phase is
formed on the surface of the water droplet. However, in each
case the acid molecules form a separate cluster either within
the water droplet (C3, C4) or on the surface of water droplet
(C5–C9). Therefore, the acid molecule self-diffusion process is
within the ‘‘bulk’’ acid materials. Interestingly, even though
the clusters formed by the acid molecules constitute only a few
hundreds of acid molecules (about 500 acid molecules), the
calculated self-diffusion coefficient reflects only the contribu-
tion from bulk diffusion. Surface diffusion is apparently not an
important factor here.
The diffusion coefficients for water molecules in different
aerosol structures were also calculated and found to be
essentially independent of the structure, with an average water
diffusion coefficient of B3.5 � 10�5 cm2 s�1 at 300 K.
7 Water vapor processing
Finally we turn our attention to how these materials might act
in the atmosphere. The accommodation of water vapor to
liquid water surfaces plays an important role in the growth of
cloud condensation nuclei into cloud droplets. Molecular
simulations of water uptake on the vapor/liquid interface of
water have been extensively studied.10,29–33 However, there are
only a few studies that have probed the effect of organic layers
on the aqueous aerosol properties.7,8 No surprising it is widely
believed that composition, concentration, structure and surface
properties of the organic coating all play a role in the water
vapor evaporation from, and condensation to such aerosols.
While some experiments have shown that surfactant films
reduce the rate of evaporation of water,34 others show for
example that butanol coatings of up to 80% monolayer
coverage do not seem to inhibit the water vapor evaporation.35
Thus considerable uncertainty exists on the effects of organic
layers on the uptake of atmospheric trace species.
In order to obtain a better understanding of the effect of
dicarboxylic acid coating on the aerosol water vapor processing,
the water sticking (mass accommodation) coefficient a was
calculated for each coated nanoaerosol system. The sticking
coefficient describing the probability of gas molecules being
incorporated into aerosol is defined as:
a ¼ number of molecules absorbed into the aerosol
number of molecules impinging the aerosol surfaceð4Þ
An accurate determination of the water sticking coefficient is
important in predicting the nucleation and growth kinetics of
cloud droplets.
To evaluate the water sticking coefficient, we placed water
monomers outside the aerosol with a random velocity pointing
to the aerosol surface, and counted the fraction at a given
thermal velocity that would be absorbed. For each simulation,
a single water molecule was placed randomly at a distance
of 5 nm from the center of the nanoaerosol, and outside the
potential cutoff distance so that initially no force was acting
between the water monomer and the coated aerosol. The water
molecule was then given an initial velocity pointing towards
the center of the aerosol which is drawn from the Maxwell–
Boltzmann speed distribution at 300 K. The trajectory of this
water monomer was monitored for 100 ps. The water monomer
was considered trapped in the aerosol if the distance between
the monomer and the coated aerosol is smaller than the size of
the aerosol, otherwise, it was considered non-absorbed. More
than 60 such water trajectories were computed for each coated
aerosol. The water sticking coefficient was then calculated as
the ratio of number of absorbed monomers to the number of
Fig. 8 Phase separation of a low surface coverage (0.6 molecules
per nm2), C9_branched acid coated nanoaerosol.
Fig. 9 Dicarboxylic acid diffusion coefficient as a function of molecular
weight. The solid line is a curve fit to a (molecular weight)�1 dependence.
This journal is c the Owner Societies 2011 Phys. Chem. Chem. Phys., 2011, 13, 9374–9384 9383
impinging monomers. The sticking coefficients were calculated
for all the structures.
Fig. 10 presents the results of sticking coefficients calcula-
tion for different coated structures. As we can see from
Fig. 10(a), the water sticking coefficient for C3_coated nano-
aerosol is essentially unity at all impinging velocities. Since we
have also seen that the C3 molecule moves to the core, the
surface of this droplet is essentially water and thus should
behave like a pure water droplet. Our result is consistent with
other researchers’ works31,33,36 on pure water droplets showing
a mass accommodation (condensation) coefficient of B1.
Fig. 10(b) shows the sticking coefficient calculation results
for C9_branched acid coated nanoaerosol. As we can see from
the figure, the sticking coefficient is the largest for incident
speeds around the most probable speed and smaller at both
lower and higher speeds. Compared with a pure water droplet,
the sticking coefficient for C9_branched acid coated nano-
aerosol is reduced, which indicates that the acid coating
impedes the mass transfer from the gas phase to the aerosol
phase. However, the C9_branched acid coated nanoaerosol
can still process water vapor. The sticking coefficients are as
high as 70% around the most probable speeds. This is due to
the apparent hydrophilic property of the surface of the coated
nanoaerosol since the hydrophilic tails of C9_branched acids
are all exposed to the environment. This water processing,
however does not lead to absorption within the water core.
Rather the processed water is associated with the hydrophilic
tail of the C9_branched acids, over the timer period of the
simulation. This is in contrast to our prior work on straight
chain fatty acids which were shown to process the water to
the core.
Because the longer chain dicarboxylic acid molecules are not
homogeneously dispersed over the droplet surface we counted
separately the water trajectories incident onto the water phase
from the water trajectories incident onto the acid phase. As
expected, the calculated sticking coefficient for water on water
was essentially unity. However, for the water molecules
impinging on the acid, almost all of them were reflected back
to the gas phase, leading to a sticking coefficient close to zero.
This result is consistent with the acid molecules forming a
thick hydrophobic layered structure.
Conclusions
In this work, molecular dynamics simulations were used
to probe the structure and the interfacial properties of
dicarboxylic acid coated aqueous aerosol. Low molecular
weight dicarboxylic acids of various chain lengths and water
solubility (from malonic acid to azelaic acid) were chosen to
coat a water droplet. The starting point of the coated aerosol is
an inverted micelle model. For malonic acid coated aerosol,
the original surface acid molecules dissolved into the water
core. For other nanoaerosols coated with low solubility acids,
phase separation between water and acid molecules was
observed during the equilibration process. The detailed phase
separation mechanism was investigated by monitoring the
structure evolution of a 10% surface acid covered nanoaerosol.
Water vapor accommodation showed that for the C3 acid
coated nanoaerosol, a water vapor accommodation factor of
1 was found for all incident water velocities. For longer chain
coated nanoaerosols, due to the surface phase separation, a
100% sticking probability was found for water monomer
colliding onto the water phase of the coated aerosol and an
almost 0% sticking probability was found for water monomer
colliding onto the acid phase.
Acknowledgements
The authors wish to acknowledge the support of a National
Science Foundation-NIRT grant and the National Institute of
Standards and Technology.
References
1 C. N. Cruz and S. N. Pandis, A study of the ability of puresecondary organic aerosol to act as cloud condensation nuclei,Atmos. Environ., 1997, 31(15), 2205–2214.
2 N. C. Shantz, W. R. Leaitch and P. F. Caffrey, Effect of organics oflow solubility on the growth rate of cloud droplets, J. Geophys.Res., 2003, 108(D5), 4168, DOI: 10.1029/2002JD002540.
Fig. 10 Water vapor sticking coefficient for different coated nanoaerosols.
9384 Phys. Chem. Chem. Phys., 2011, 13, 9374–9384 This journal is c the Owner Societies 2011
3 B. Ervens, G. Feingold, S. L. Clegg and S. M. Kreidenweis, Amodeling study of aqueous production of dicarboxylic acids: 2.Implications for cloud microphysics, J. Geophys. Res., 2004,109(D15), D15206, DOI: 10.1029/2004JD004575.
4 D. J. Donaldson, H. Tervahattu, A. F. Tuck and V. Vaida,Organic aerosols and the origin of life: An hypothesis, Origins LifeEvol. Biospheres, 2004, 34(1/2), 57–67.
5 C. M. Dobson, G. B. Ellison, A. F. Tuck and V. Vaida, Atmosphericaerosols as prebiotic chemical reactors, Proc. Natl. Acad. Sci. U. S. A.,2000, 97(22), 11864–11868.
6 D. J. Donaldson and V. Vaida, The influence of organic films at theair-aqueous boundary on atmospheric processes, Chem. Rev.,2006, 106(4), 1445–1461.
7 P. Chakraborty and M. R. Zachariah, Sticking coefficient andprocessing of water vapor on organic-coated nanoaerosols,J. Phys. Chem. A, 2008, 112(5), 966–972.
8 P. Chakraborty and M. R. Zachariah, ‘‘Effective’’ negative surfacetension: A property of coated nanoaerosols relevant to the atmosphere,J. Phys. Chem. A, 2007, 111(25), 5459–5464.
9 G. B. Ellison, A. F. Tuck and V. Vaida, Atmospheric processing oforganic aerosols, J. Geophys. Res., 1999, 104(D9), 11633–11641.
10 C. E. Kolb, R. A. Cox, J. P. D. Abbatt, M. Ammann, E. J. Davis,D. J. Donaldson, B. C. Garrett, C. George, P. T. Griffiths, D. R.Hanson, M. Kulmala, G. McFiggans, U. Poschl, I. Riipinen,M. J. Rossi, Y. Rudich, P. E. Wagner, P. M. Winkler,D. R. Worsnop and C. D. O’Dowd, An overview of current issuesin the uptake of atmospheric trace gases by aerosols and clouds,Atmos. Chem. Phys., 2010, 10(21), 10561–10605.
11 M. Hori, S. Ohta, N. Murao and S. Yamagata, Activationcapability of water soluble organic substances as CCN,J. Aerosol Sci., 2003, 34(4), 419–448.
12 K. Kawamura, R. Semere, Y. Imai, Y. Fujii and M. Hayashi,Water soluble dicarboxylic acids and related compounds inAntarctic aerosols, J. Geophys. Res., 1996, 101(D13), 18721–18728.
13 M. Z. H. Rozaini and P. Brimblecombe, The odd–even behaviourof dicarboxylic acids solubility in the atmospheric aerosols, Water,Air, Soil Pollut., 2008, 198(1–4), 65–75.
14 M. Bilde, B. Svenningsson, J. Monster and T. Rosenorn,Even–odd alternation of evaporation rates and vapor pressuresof C3–C9 dicarboxylic acid aerosols, Environ. Sci. Technol., 2003,37(7), 1371–1378.
15 T. M. Raymond and S. N. Pandis, Cloud activation of single-component organic aerosol particles, J. Geophys. Res., 2002,107(D24), 4787, DOI: 10.1029/2002JD002159.
16 S. A. Safran, Statistical Thermodynamics of Surfaces, Interfaces,and Membranes, Addison-Wesley Pub., Reading, Mass, 1994.
17 A. H. Fuhrhop and T. Y. Wang, Bolaamphiphiles, Chem. Rev.,2004, 104(6), 2901–2937.
18 C. E. Corrigan and T. Novakov, Cloud condensation nucleusactivity of organic compounds: a laboratory study, Atmos. Environ.,1999, 33(17), 2661–2668.
19 S. Plimpton, Fast parallel algorithms for short-range molecular-dynamics, J. Comput. Phys., 1995, 117(1), 1–19.
20 H. J. C. Berendsen, J. R. Grigera and T. P. Straatsma, The missingterm in effective pair potentials, J. Phys. Chem., 1987, 91(24),6269–6271.
21 H. C. Berendsen, J. P. M. Postma, W. F. van Gunsteren andJ. Hermans, Interaction Models for Water in Relation to ProteinHydration, in Intermolecular Forces, ed. B. Pullman, Reidel,Dordrecht, The Netherlands, 1981, pp. 331–342.
22 B. C. Garrett, G. K. Schenter and A. Morita, Molecular simulationsof the transport of molecules across the liquid/vapor interface ofwater, Chem. Rev., 2006, 106(4), 1355–1374.
23 J. Alejandre, D. J. Tildesley and G. A. Chapela, Molecular-dynamics simulation of the orthobaric densities and surface-tension of water, J. Chem. Phys., 1995, 102(11), 4574–4583.
24 J. P. Ryckaert, G. Ciccotti and H. J. C. Berendsen, Numerical-integration of Cartesian equations of motion of a system withconstraints-molecular-dynamics of N-alkanes, J. Comput. Phys.,1977, 23(3), 327–341.
25 D. Chandler, Interfaces and the driving force of hydrophobicassembly, Nature, 2005, 437, 640–647.
26 D. Chandler, Hydrophobicity: Two faces of water, Nature, 2002,417(6888), 491–491.
27 K. Lum, D. Chandler and J. D. Weeks, Hydrophobicity at smalland large length scales, J. Phys. Chem. B, 1999, 103(22), 4570–4577.
28 G. Strobl, The physics of polymers-concepts for understanding theirstructures and behavior, Springer, 1996.
29 G. Nagayama and T. Tsuruta, A general expression for the conden-sation coefficient based on transition state theory and moleculardynamics simulation, J. Chem. Phys., 2003, 118(3), 1392–1399.
30 L. X. Dang and B. C. Garrett, Molecular mechanism of water andammonia uptake by the liquid/vapor interface of water, Chem.Phys. Lett., 2004, 385(3–4), 309–313.
31 A. Morita, M. Sugiyama, H. Kameda, S. Koda and D. R. Hanson,Mass accommodation coefficient of water: Molecular dynamicssimulation and revised analysis of droplet train/flow reactor experi-ment, J. Phys. Chem. B, 2004, 108(26), 9111–9120.
32 R. Vacha, P. Slavicek, M. Mucha, B. J. Finlayson-Pitts andP. Jungwirth, Adsorption of atmospherically relevant gases atthe air/water interface: Free energy profiles of aqueous solvationof N-2, O-2, O-3, OH, H2O, HO2, and H2O2, J. Phys. Chem. A,2004, 108(52), 11573–11579.
33 J. Vieceli, M. Roeselova and D. J. Tobias, Accommodationcoefficients for water vapor at the air/water interface, Chem. Phys.Lett., 2004, 393(1–3), 249–255.
34 J. Buajarern, L. Mitchem and J. P. Reid, Manipulation andcharacterization of aqueous sodium dodecyl sulfate/sodiumchloride aerosol particles, J. Phys. Chem. A, 2007, 111(50),13038–13045.
35 J. R. Lawrence, S. V. Glass and G. M. Nathanson, Evaporation ofwater through butanol films at the surface of supercooled sulfuricacid, J. Phys. Chem. A, 2005, 109(33), 7449–7457.
36 T. Tsuruta and G. Nagayama, Molecular dynamics studies on thecondensation coefficient of water, J. Phys. Chem. B, 2004, 108(5),1736–1743.
